Background
Evidence suggests that the archetypal pro-inflammatory cytokine, interleukin-1beta (IL-1β), is a key component in the pathogenesis of stroke, Alzheimer’s disease (AD) [for review, see
1] and Parkinson’s Disease (PD) [for review, see
2]. These are complex pathologies in which it can be difficult to determine the physiological actions of IL-1β in the brain. However, by examining the consequences of a cerebral IL-1β challenge we may facilitate our understanding of the role of this cytokine in neurodegenerative disease and disorder.
Intrastriatal microinjection of IL-1β in three-week-old (P21) juvenile rats induces a chronic reduction in the apparent diffusion coefficient (ADC) of tissue water on magnetic resonance imaging (MRI) [
3]. This is indicative of a loss of high-energy phosphates, detectable using non-invasive phosphorus (
31P) magnetic resonance spectroscopy (MRS), in experimental acute cerebral ischaemia [
4‐
6]. Mechanistically, a reduction in cerebral blood flow (CBF) promotes the failure of adenosine triphosphate (ATP)-dependent cell membrane ion pumps, dysregulating cell volume [for review, see [
7], inducing cell swelling, reducing the extracellular space and, therefore, the ADC [
8‐
10].
Contrary to classic ischaemia where ADC changes occur swiftly within a matter of minutes [
11], IL-1β-induced changes do not present until 6 h post challenge [
3]. Furthermore, IL-1β increases regional cerebral blood volume (rCBV), lactate levels remain unaltered [
3], and there is an absence of neuronal cell death [
12,
13]. It has been suggested that IL-1β-induced neutrophil-mediated blood-brain-barrier (BBB) breakdown may influence water diffusion within the rat brain parenchyma, affecting ADC changes [
3]. However, BBB breakdown occurs between 4 h and 5 h and gadolinium post-contrast enhancement diminishes by 6 h when ADC first becomes significantly reduced [
3]. Furthermore, the BBB reseals itself by 24 h whereas ADC changes persist up to 123 h, and are not prevented by neutrophil depletion [
3]. IL-1β-induced morphological changes such as microglial activation or cell swelling have also been suggested as possible explanations for the reduced ADC [
12]. Electrical impedence measurements in a rat model of NMDA-induced excitotoxic injury have shown that an ~10 % cell swelling corresponds to an ~50 % reduction in ADC [
10]. However, microglia revert to resting morphology 72 h post IL-1β challenge whilst ADC remains depressed beyond this time point [
3]. Thus, although ischaemia, BBB changes and cell swelling do not seem to account for the ADC reduction observed in this model, IL-1β-induced energy failure in vivo remains a possibility.
Reduced ADC is observed in rat models of
N-methyl-D-aspartate (NMDA)- [
14] or ouabain-induced [
15] excitotoxic injury, and bicuculline-induced status epilepticus [
16], which exhibit no apparent ischaemic component to their pathology. Nevertheless, these studies sought to correlate ADC changes with the energy status of the brain given the strong association between failure of energy-dependent transmembrane Na
+/K
+-ATPase ion pumps and dysregulation of cell volume [for review, see
7]. Therefore, the metabolic status of brain parenchyma following IL-1β challenge in vivo warrants examination to determine whether an ischaemia-independent energy deficit may be associated with the reduction in ADC.
IL-1β does not appear to affect high-energy phosphate metabolism of an
ex vivo preparation of brain parenchyma
per se [
17].
31P MRS was used to characterise the energy status of organotypic hippocampal-slice cultures (OHSCs) prior to and following IL-1β challenge. The OHSC set-up preserves brain parenchyma cytoarchitecture and synaptic connectivity without the complication of a functional vascular component. Thus, the vascular-related consequences of an IL-1β challenge in vivo such as neutrophil recruitment, blood–brain barrier (BBB) and perfusion changes [
3] can be extricated from those pertaining to the parenchyma.
The question of whether IL-1β affects cerebral energy metabolism remains partly addressed; tissue energy status should additionally be addressed in vivo where the physiological consequences of a functional vasculature are included. Furthermore, the classic bio-imaging marker of an energy deficit - a reduced ADC - may only be observed, followed and correlated with metabolism in real-time, in vivo.
Herein, the energetic contribution to the IL-1β-induced reduction in ADC in vivo was examined. A comprehensive characterisation of the real-time longitudinal evolution of the lesion on ADC MRI is presented. Localised
31P MRS-detectable energy status, as determined by the phosphocreatine (PCr) to ATP ratio [
18], was measured within the ipsilateral and contralateral striatum 6 h following intrastriatal microinjection of 1 ng/μl IL-1β when ADC first becomes reduced at this dose [
3]. Additionally,
31P MRS, ADC and anatomical [T
1-weighted (T
1-w)/T
2-weighted (T
2-w)] MRI data were acquired serially in the same animal at 2.5 h, 6 h, 24 h and 72 h following an intrastriatal 100 ng/μl IL-1β challenge.
Discussion
In the present study, the energetic contribution to the IL-1β-induced reduction in ADC was examined in vivo, for the first time. A potential energy deficit was hypothesised since failure of the ATP-dependent transmembrane ion pumps [for review, see
7] is critical in controlling the volume of the intra- and extracellular space, and thus the ADC [
8‐
10].
An IL-1β-induced (100 ng/μl) reduction in ADC was observed from 6 h to 72 h post challenge (as determined by thresholding), and was due to a combination of the ADC of the initial ROI at 6 h (ROI
6 h) remaining depressed throughout the study (Fig.
2b) and the ADC of the spatial difference between ROI 6 h and ROI 24 h (ROI
6 h-24 h) becoming further depressed (Fig.
2c). However, the ADC within the dimensions of the
31P MRS voxel was reduced only at 24 h (Fig.
1d), coinciding with the greatest spatial extent of ADC change (Fig.
1c), and returned to baseline by 72 h (Fig.
1d). The difference in sensitivity with respect to both analyses is important, as the ADC changes detected by thresholding are spatially smaller than the dimensions of the
31P MRS voxel (Fig.
1a). If an IL-1β-induced reduction in ADC is indeed associated with energy failure, the spatial extent of ADC changes at 6 h and 72 h may not be sufficient to correlate with
31P MRS-detectable perturbations in the PCr to γATP ratio. However, the large spatial extent of ADC change at 24 h (Fig.
1a and
c) in parallel with a significantly reduced ADC (Fig.
1b and
d) suggests that one should, at least, be able to detect energy failure at this time point.
Interestingly,
31P MRS did not detect a change in the PCr to γATP ratio at any time point following 100 ng/μl IL-1β challenge (Fig.
4b), nor at 6 h following 1 ng/μl IL-1β challenge (Fig.
4a) when ADC is first reported to become reduced at this dose [
3]. This finding is in agreement with previous observations using an
ex vivo preparation of rat brain parenchyma that demonstrated an unperturbed
31P MRS-detectable energy status 6 h following incubation with IL-1β [
17].
In vivo localised rat brain 31P MRS is a technically challenging method and was used in the present study to measure the tissue energy status specifically within the area of IL-1β microinjection, minimising contamination of signal from surrounding non-injected tissue. However, this approach has sensitivity limitations. The dimensions of the 31P MRS voxel must be sufficiently large for signal detection within a reasonable time acquisition (in this case, 1 h). In the present study, the mean peak height for PCr within the left and right cerebral hemispheres for 3 control animals was 85 mm, and the standard deviation 19 mm. Thus, within the limits of sensitivity of this protocol, a minimal reduction of approximately 25 % in PCr is required to detect a change in the PCr to γATP ratio.
One should consider how large a reduction in the PCr to γATP ratio might be required to constitute energy failure with confidence. In cerebral ischaemia in the rat, in vivo PCr and ATP levels measured by
31P MRS have been shown to decrease in parallel [
26], in which case one would not be able to detect an overall change in the the PCr to γATP ratio. However, these observations were based on extreme ends of the spectrum where the metabolites were either nearing control values or were extremely compromised [
27‐
29]. Moderate ischaemic challenges where
31P MRS-detectable PCr is reduced but ATP is maintained have demonstrated that PCr must decrease to 40 ± 26 % before ATP loss occurs [
30]. When PCr is above 25 % of control value, ATP loss is one-third that of PCr whereas below 25 % of control PCr value, ATP loss supersedes that of PCr [
30].
A better estimate of the sensitivity of the
31P MRS protocol may be provided by a power calculation. The sample size required to detect a 25 % change in the PCr to γATP ratio 24 h following 100 ng/μl IL-1β challenge, the time point exhibiting the spatially greatest and severest reduction in ADC, is
n = 49. This is impractical given the challenging nature and long duration of the
31P MRS protocol, and reinforces the unlikelihood that IL-1β induces energy failure in vivo. By comparison, in rat cerebral ischaemia, whilst PCr is above 25 % of control value, ATP loss is one-third that of PCr [
30], as explained earlier. If, on that basis, a power calculation is performed to show how many animals are required to demonstrate a 75 % change in the PCr to γATP ratio,
n = 5. Therefore, within the sensitivity limitations of the
31P MRS protocol, it is likely that a change in the PCr to γATP ratio would only be detectable if the metabolic impact was severe enough, which does not appear to be the case following IL-1β challenge.
It is important to address the correlation between ADC reduction and energy failure using in vivo models. In the classic middle cerebral artery occlusion (MCAO) model of brain ischaemia, energy failure perturbs ionic gradients inducing cell oedema and reducing the extracellular space, which depresses the ADC to 60–70 % of the original value by 7 h [
6,
31‐
33]. The association between ADC changes and metabolic failure following MCAO has been proven by anatomic mapping of regions of depressed ADC on MRI with areas of acidosis, lactate accumulation [
34] and ATP loss, assessed using bioluminescence imaging on frozen tissue [
6]. One may conclude that a decrease in ADC of approximately 40 % of its control value or the value in the non-injected contralateral hemisphere constitutes energy failure in MCAO.
The same reasoning cannot be translated to other non-ischaemic examples of ADC reduction in vivo. For example, an excitotoxic NMDA challenge in the rat brain induces a comparable ~ 40 % reduction in ADC, which affects almost the entire injected hemisphere, without tissue energy failure as assessed by
31P MRS [
14]. Injection of the excitotoxin, ouabain, into the rat brain produces an approximate 40 % drop in ADC, albeit within a smaller sized lesion but without
31P MRS-detectable metabolic failure [
15]. Interestingly, brain ADC reductions of ~ 18 % have been reported in an experimental model of status epilepticus, which did not correlate with energy failure [
16]. Thus, given our present incomplete understanding of the mechanisms of in vivo ADC reduction, it is not possible to make a direct association between magnitude of ADC reduction and prediction of tissue energy failure in models other than MCAO.
Partial volume effects are significant with respect to both ADC and
31P MRS measurements. Following IL-1β challenge, the partial volume effects on ADC measurement have been demonstrated by comparing the ROI analysis with the
31P MRS voxel analysis (Fig.
1b and
d, respectively). In Fig.
1b, the ADC at 24 h is reduced within the injected hemisphere to ~ 90 % of the value in the contralateral hemisphere. In Fig.
1d, the ADC measurement within the confines of the
31P MRS voxel detects an ADC reduction to 95 % of the value in the contralateral hemisphere, thus halving the magnitude. IL-1β diffusibility may appear unrestricted since ADC changes were observed throughout the cerebral hemisphere, particularly at 24 h, however, it seems that only 50 % of the parenchyma contained within the
31P MRS voxel was affected by tissue water diffusion changes, which would be expected to affect the sensitivity of
31P MRS-detectable changes. However, as discussed earlier, one cannot draw a direct correlation between magnitude of ADC reduction and energy failure in non-ischaemic challenges.
It is of note that astrocytes are the only brain cell-type to exhibit a glycogen reserve, utilised during ischaemia to preserve neuronal viability [
35]. Interestingly, challenging dissociated primary neuronal or astrocytic cell cultures with IL-1β induces astrocytic utilisation of their endogenous glycogen reserve whilst neuronal metabolism is unaltered [
36,
37]. Hence, an IL-1β challenge in vivo may not alter the overall PCr to γATP ratio of brain parenchyma.
The present study has shown that the ADC reduction cannot be confidently attributed to depleted ATP levels, as measured by
31P MRS. This finding has important implications for the interpretation of ADC changes in the brain because it occurs under metabolic conditions distinct from those concerning classic cerebral ischaemia, as is the case with rodent models of excitotoxic injury [
14,
15] and status epilepticus [
16]. There are clear differences in the consequences of an ischaemic, excitotoxic or IL-1β-induced challenge for neurons. Ischaemic or excitotoxic challenge induces neuronal cell death [
14,
15] whereas an intrastriatal IL-1β microinjection is not neurotoxic [
12,
13]. Therefore, any mechanistic explanation for reduced ADC following IL-1β challenge is likely to be different compared to that following excitotoxic injury or ischaemia.
An improved understanding of the mechanisms underlying brain ADC changes in non-ischaemic challenge and a closer examination of non-ischaemic in vivo models of ADC reduction are warranted as this may facilitate an improved understanding of brain pathophysiology observed non-invasively by routine ADC MRI. This is particularly pertinent at the present time as IL-1β appears to be a key player in the pathogenesis of neurodegenerative diseases such as Alzheimer’s Disease and Parkinson’s Disease [for reviews, see
1,
2]. There is great interest in MR examination of these diseases clinically and pre-clinically, particularly with respect to tissue water diffusion measurements [
38‐
40]. In order to interpret the MR results effectively, it would be useful to have an improved understanding of the contribution of the IL-1β-induced component of the pathology to the brain tissue water diffusion signal.
Competing interests
The author declares no competing interests.
The MRC Biochemical and Clinical Magnetic Resonance Unit (University of Oxford) was closed down during the course of the author’s D.Phil. studies.